A sound idea

How Live Transcribe went from helping a team to communicate — to helping millions of people

3-minute read

Google research scientist Dimitri Kanevsky sits on a blue couch conversing with colleagues. From behind his right shoulder, his left hand is seen holding a phone with text-to-speech captions visible on the screen.

“What jump-started Live Transcribe was one person caring about another person in the company and doing something about it.”

Eve Andersson, Accessibility & Disability Inclusion Director

Hear from Chet, Dimitri, Sagar, and other Googlers on how the team came together to launch Live Transcribe

Watch the Video


Produced in partnership with ATTN:, a media company creating purpose-driven stories

After decades of creating innovative solutions to communicate, Dimitri Kanevsky, who lost his hearing at an early age, worked with his Google teammates to create Live Transcribe — a speech-to-text mobile app that helps him engage with spoken words and surrounding sounds in real time. Today, after years of testing and refinement in collaboration with the deaf and hard of hearing community, this technology enables millions of people to be a part of every conversation.

Dimitri Kanevsky faces the camera sitting down and smiling. He has short gray hair and is wearing a purple shirt.

Dimitri Kanevsky, a Speech Research Scientist at Google

Bridging the communication gap

“When Chet developed the prototype ... I told him, ‘I’ve been dreaming about this my whole life!’”

Dimitri Kanevsky, Speech Research Scientist

As a research scientist working to improve speech recognition accuracy, Dimitri joined Google in 2014. In meetings with colleagues, he used CART, a professional interpreter service that displays speech-to-text captions in real time on a dedicated monitor. Although it was helpful, CART required multiple devices and advance preparation. Communication with his team members — including engineer Chet Gnegy and product manager Sagar Savla — also happened through more improvised methods: using note-taking apps, passing sticky notes, even hand gestures.

This experience led Chet to test an idea. He knew that speech transcription accuracy had advanced significantly, thanks in large part to Dimitri’s contributions to the field. But was the technology good enough to capture and display conversations on a phone’s screen in real time? He built a rough prototype and gave it to Dimitri to pilot. “When Chet developed [it], there were a lot of transcript errors,” Dimitri recalls. “He would ask, ‘How can you use this?’ And I told him, ‘Are you kidding? I’ve been dreaming about this my whole life!’”

Dimitri using Live Transcribe during a video call with his family

Broadening the conversation

“There are millions of people in the world who are deaf — most who do not communicate in English or have means to use expensive captioning services. We had to find a way to not only make the technology available in many languages, but also to make it free.”

Sagar Savla, Product Manager for speech recognition products

Seeking additional input, Dimitri, Sagar, and Chet brought the prototype to an accessibility innovation sprint, where Google teams from around the world pitch new ideas and exchange feedback on accessibility products. After receiving enthusiastic internal support, Sagar knew that the app had the potential to help millions of people — including his grandmother, who is hard of hearing. With the help of Gallaudet University — the world’s foremost institution for the education of the deaf and hard of hearing — he led the team to turn the prototype into a publicly available product.

Project Euphonia is a Google Research initiative focused on helping people with atypical speech be better understood.
See how you can assist our research
Sagar Savla speaks with a Gallaudet University sign behind him. Standing next to him is a female sign language interpreter.

Sagar during a visit to Gallaudet University in Washington, D.C.

Live Transcribe launched in 2019, transcribing real-time speech in over 70 languages on Android and Chrome OS devices. A year later, the app was updated to also include notifications that alert users of critical sounds in one’s environment — a feature that helps not only people who are deaf or hard of hearing but also those who are unable to hear noises temporarily, such as when someone is wearing headphones.

The ideas don’t stop there: future enhancements include adding even more languages, increased transcription accuracy, and better experiences for those communicating across languages or in group settings. Downloaded over 100 million times as of 2021, Live Transcribe underscores the immense impact a single idea can have toward creating richer, more inclusive human connections.

“For the first time, I could speak with my granddaughters. It was amazing to talk to them, to play chess, to hear their stories.”

Dimitri Kanevsky, Speech Research Scientist

More on this topic

Related Stories